Sorting of records with duplicate removal in a database system
    4.
    发明授权
    Sorting of records with duplicate removal in a database system 有权
    在数据库系统中对重复删除的记录进行排序

    公开(公告)号:US07370068B1

    公开(公告)日:2008-05-06

    申请号:US10234923

    申请日:2002-09-04

    申请人: Son Pham Thu K. Pham

    发明人: Son Pham Thu K. Pham

    IPC分类号: G06F12/00 G06F17/30

    摘要: A database system includes an enhanced technique for performing sorts in which removals of duplicate records are performed to compact the size of segments being sorted. To enhance a query plan, the duplicate record removal is performed as early in the query plan as possible. By removing duplicate records early in the query plan, the number of input/output (I/O) operations is reduced, resulting in more efficient usage of database system resources. In example implementations, two type of sorts are performed: a heap sort (to sort successive segments of an input file, with the sorting associated with concurrent removal of duplicate records to compact each segment so that a smaller number of I/O accesses is needed); and a merge sort (in which output files from prior sorting passes are merged and sorted, with the merge sort process also associated with the removal of duplicate records to further compact the data segments and reduce the number of I/O accesses).

    摘要翻译: 数据库系统包括用于执行排序的增强技术,其中执行重复记录的删除以压缩被排序的段的大小。 为了增强查询计划,尽可能早地在查询计划中执行重复记录删除。 通过在查询计划中提前删除重复记录,减少输入/输出(I / O)操作的数量,从而更有效地使用数据库系统资源。 在示例实现中,执行两种类型的排序:堆排序(排序输入文件的连续段,排序与并发删除重复记录相关联以压缩每个段,从而需要较少数量的I / O访问 ); 和合并排序(其中来自先前排序遍历的输出文件被合并和排序,合并排序过程也与删除重复记录相关联,以进一步压缩数据段并减少I / O访问的数量)。

    Providing a join plan using group-by operator
    8.
    发明授权
    Providing a join plan using group-by operator 有权
    使用分组运算符提供连接计划

    公开(公告)号:US06757677B2

    公开(公告)日:2004-06-29

    申请号:US09967561

    申请日:2001-09-28

    申请人: Son Pham Thu K. Pham

    发明人: Son Pham Thu K. Pham

    IPC分类号: G06F1730

    摘要: A database system includes an enhanced optimizer module that is able to consider group-by operations as part of the optimization of a join. Given a query that performs a join of multiple tables and that includes an aggregation function as well as at one of a Where clause and a Group-by clause, a partial group-by is performed on at least one of the tables of the join query to reduce the number of rows for the join. Group-by can also be performed on intermediate results of intermediate joins.

    摘要翻译: 数据库系统包括增强优化器模块,该模块能够将分组运算视为优化连接的一部分。 给定执行多个表的连接并且包括聚合函数以及Where子句和Group-by子句之一的查询,对连接查询的至少一个表执行部分组合 以减少连接的行数。 也可以对中间连接的中间结果进行分组。

    Load balancing in a network with session information
    10.
    发明授权
    Load balancing in a network with session information 有权
    在具有会话信息的网络中进行负载平衡

    公开(公告)号:US08776207B2

    公开(公告)日:2014-07-08

    申请号:US13356399

    申请日:2012-01-23

    摘要: Methods and systems for balancing load among firewall security devices are provided. According to one embodiment, a switch maintains a session table the session entries of which represent established traffic sessions between a source and a destination and form an association between the traffic session and a particular firewall security device (FSD). Responsive to receiving a packet of a first traffic session on a first port, a determination is made whether there exists a matching session entry. Responsive to a negative determination, a load balancing function is performed to select an FSD with which to associate the first traffic session and a corresponding reverse second traffic session. After processing of the packet by the selected FSD and receipt of the packet at a second port, a session entry is installed within the session table for the second traffic session and which associates the selected FSD with the second traffic session.

    摘要翻译: 提供了在防火墙安全设备之间平衡负载的方法和系统。 根据一个实施例,交换机维护会话表,其会话表项表示源和目的地之间建立的业务会话,并形成业务会话与特定防火墙安全设备(FSD)之间的关联。 响应于在第一端口上接收到第一通信会话的分组,确定是否存在匹配的会话条目。 响应于否定确定,执行负载平衡功能以选择与第一业务会话关联的FSD和相应的反向第二业务会话。 在通过所选择的FSD处理分组并在第二端口接收到分组之后,在第二业务会话的会话表中安装会话条目,并且将所选择的FSD与第二业务会话相关联。